Triagem virtual e biológica de substâncias com potencial atividade antiviral contra três flavivírus: dengue virus, yellow fever virus e Zika virus

Abstract

Approximately three billion people live in regions at risk of infection by the dengue virus (DENV), Zika virus (ZIKV) and yellow fever virus (YFV). In addition to severe cases of dengue, Zika presents complications such as congenital syndromes in fetuses. Yellow fever outbreaks also happened in Brazil in 2017, even with a vaccine available against it. Currently, there are no licensed antivirals for the treatment of these diseases, emphasizing the need for strategies to obtain new bioactive molecules with antiviral activity. The objective of this work was to evaluate, through in vitro and in silico strategies, the antiviral activity of substances against DENV, ZIKV and YFV, targeting the viral protease (NS3PRO). Three-dimensional structures of peptidomimetic compounds inhibiting DENV-3 protease were used to generate models of the quantitative relationship between chemical structure and biological activity by holograms (HQSAR) and molecular docking. Binding sites and hotspots of protein structures (PDB: 3U1I and 5YOF) were detected using the FTSite, FTMap and AutoGrid programs. These results were used to the construction of a pharmacophoric model for virtual screening. Physical and chemical properties, presence molecular fragments and interactions with the molecular target (pharmacophore) were defined in the screening. Eight compounds were selected after molecular dynamics, as well as predictions of toxicity and biological activity. Biological activity was confirmed by the colorimetric assay of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), in order to obtain the cytotoxic (CC50) and effective (EC50) concentrations, as well as selectivity indixes (SI). The antiviral activity was also evaluated by plaque reduction, pre-treatment and virucide activity assays. Initially, HQSAR models were built and validated, with an optimal model (q2 = 0.675 and r2 = 0.873) selected using chemical contribution maps of molecular fragments. The catalytic triad, ligands and biding sites pointed at the structures were used to build a pharmacophoric model based on mapping, interpretation of the HQSAR model and interactions observed in previous fitting studies. A virtual screening of approximately 7,600,000 compounds was conducted in several stages (pharmacophore, docking and molecular dynamics), identifying eight potential inhibitors to the DENV-3 protease (two drugs and six synthetic molecules under investigation). Molecular dynamics simulations showed stable interactions at the target sites, and were complemented by favorable biological activity (5/8) and toxicity (8/8) predictions. Among the substances, one was active against ZIKV (EC50 = 11.86 ± 0.43 μM; IS = 1.42), four against YFV (EC50 = 4.79 ± 0.17 to 24.15 ± 1.15 μM; IS = 1.97 to 3.74), two against DENV- 2 (EC50 = 4.21 ± 0.14 and 9.51 ± 0.12 μM; 2.26 and 3.16) and one against DENV-3 (EC50 = 37.51 ± 0.8 μM; IS = 2.85). In plaque reduction assays, substances 136 and 128 reduced about 1.0 to 1.5 log10 of the viral titer (PFU/mL) of ZIKV, YFV and DENV-2. Substances 136 and 140 also reduced about 1.0 log10 of YFV titer in pre-treatment assays. As a perspective, enzyme inhibition assays will be carried out in order to confirm the molecular target predicted in this work.